Iris recognition based on 2D rotation invariant feature

• Main Authors: Chun - Hui Lin, 林俊輝
• Other Authors: Yao-Hong Tsai
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/85694463277256198439

碩士 === 玄奘大學 === 資訊管理學系碩士班 === 103 === For iris recognition, it will result in recognition errors or low recognition rate if the eye image was captured under rotation or displacement in the image plane. To deal with the problem of image rotation, this paper combines the features of rotational invariance to solve the problem of low recognition rate by using the local binary pattern (LBP) that preserves the regional characteristics of iris images. LBP is usually used to describe changes of the texture patterns in images. The main advantage of LBP is its simple operation and the characteristics of avoiding shadow effects such that it is suitable for real-time systems. The rotational invariance is characterized by a unified method to reduce the dimension of rotated features of iris images and the coding of rotational invariance also reduce the degree of difference between iris features. Finally, the captured iris feature combines the weighted value using an iris mask in order to improve the total recognition rate.